Cuernavaca 2006 r2-3

Dont grasp why Volo played Ng7 instead of Qxc3.
In any case very nice to see the ressurrexion of a famous, historical line from games that made top players stop playing it as white. In 2006 Naka wins with it! Did you hear that Vassily Smyslov?

The extra pawn is not worth much, and I guess that white would get heavy pressure in the a and b-files after 21.Rb3. Black has problems developing his knight, and White has a possibility of Nxf5 in some variations.

Great to see Hikaru win a good one. he has now drawn #1 seed Ponomariov in Round 1. and won his game against #2 seed Volokitin. this is a very impressive start in my opinion. I now think he has a good shot at winning the tournament. I will be here tonight at 5 pm to again watch all the excitement. Go Hikaru.

I'm not surprised at Hikaru's win.

I did not believe Qxc3 would have been good because it would then free Hikaru's dark-squared bishop to operate on the a1-h8 diagonal. He would play Bc1 and then at some point Bb2. With the pawn on c3, Hikaru's bishop was stifled.

What is the time control for play in Mexico?

What are the Sofia rules mentioned?

THANKS

Details in the Cuernavaca item two days ago. "The time control will be 100 minutes for the entire game, plus 30 seconds increment per move."

Sofia/Mtel rules prohibit agreeing to a draw without consulting an arbiter and/or committee to ensure the position is clearly drawn. Basically it's legislation to prevent short draws.

In a recent interview of GM Svidler, one of the legends of Russian chess (being multiple champion), he wished FIDE to stop meddling with the time controls, which are being changed often enough.

As I notice, though, many chess organizers apply different time controls, and even rules!? So, I don't really get what's the preffered time control for chess, recommended (and implemented)by the majority?!

I personally prefer adjourning games after 40 or 60 moves, as it was done 20 years ago. Adjournment was banned to the joy of most arbiters, I remember.

Computer help after adjournment? I suppose Intel, Toshiba, ChessBase, Convekta, Chess Informant and others would welcome it!?

http://www.uaem.mx/ajedrez2006/ronda3/tfd.htm

In IM Leon - GM Nakamura 0-1, Hikaru reminded me of Topalov. Cat and mouse play.

And I like the viewer very much.

Volokitin played in the Greek Team Championship last year. He didn't leave a good impression there. Is Nakamura playing for any European team, or only in US?

Last post today.

Where can I find a chess biography on this interesting GM Nakamura?

He slapped every opening chess book ever written (not to mention some GMS, too), in the face, by promoting successfuly 1.e4 e5 2.Qh5?

Will all chess opening books have to be rewritten, after Nakamura, I wonder?

Not even Morozevich came close, with his 1.e4 c5 2.Na3?

Nakamura's chess biography may give a lead to mass producing new grandmasters of his modern type. His playing style is very enjoyable.

IM Petronic,

As you've noticed, bios of chessplayers are often difficult to find. I don't believe GM Nakamura has his own website. (Note to any GM who hopes to make money through endorsements or other fan support related businesses: people need to be able to find you and find out about you!)

Here are a few things that are available:
1. Official US Chess Federation bio. Rather out of date (it doesn't list him as US Champion). I believe it was last updated in 2004.

http://www.uschess.org/news/bio/nakamura.php

2. Wikipedia bio. Not bad.

http://en.wikipedia.org/wiki/Hikaru_Nakamura

3. ChessCafe interview with GM Nakamura One of the more interesting interviews. Available only in .pdf form, which means it doesn't get indexed well at Web search engines:

http://www.chesscafe.com/text/skittles242.pdf

ChessDrum also has an html version of the same interview:

http://www.thechessdrum.net/newsbriefs/2004/NB_Nakamura.html

4. 2004 bio from the 2005 US Championship

http://www.uschesschampionship.com/news/006nakamura.htm

Chess books rewritten? Why? Hikaru gave no reason to rewrite any chess books. Before he started his 2.Qh5 adventures, this move was considered to give white equality-at-best, and his exploits changed nothing. White still has equality-at-best. Nakamura is a middlegame wizard, of course, but the rumors about him being the death of the opening theory are greatly exaggerated.

What was the middlegame before, is the opening now.

Nakamura is beating GMs starting with opening moves that are not recommended *by GMs" in any serious modern theory literature.

If something like this is happening, one has to wonder whether the existing theory advocates the best lines, or maybe, not? I would not be surprised if all existing opening theory, is, in fact, based on unstable foundations, given the results from practice.

Even Fischer may be surprised by these developments and give up on his "random chess" idea.

When anyone (either Hikaru or someone else) starts regularly getting favorable positions from the 2.Qh5 opening against 2730+ GMs, then you may start talking about AMENDING books, not rewriting them. Chances are this is never going to happen.

Amending is not what I had in mind.
What are your reasons 1.e4 e5 and, for example, 2.Nf3 (recommended) should be favored over 1.e4 e5 2.Qh5?

"If something like this is happening, one has to wonder whether the existing theory advocates the best lines, or maybe, not? I would not be surprised if all existing opening theory, is, in fact, based on unstable foundations, given the results from practice."

Jeez. It was ONE game. Overreact much? Anand could probably destroy everyone posting here with 1.Nf3 and 2.Ng1. Doesn't mean its suddenly sound opening theory.

IMJP.ST. "Last post today."

What was the Over/Under on this promise? 2.5?

Speaking of which: Seahawks +4, Under 47. GO HAWKS.

It's almost midnight over here. The historical importance of the topic overrides the promise. And - my kid went to sleep earlier than expected.

My point is that chess (and chess opening theory)is far from explored, and - Nakamura realized this, just in time, when many were about to change the Rules of Chess. If you read Russian, Mark Dvoretsky just made a post about the vast possibilities in chess at http://www.chesspro.ru.

About Anand's future opening moves, losing two tempi as White, without creating any weaknesses, in current theory, does not deteriorate White's position enough to promise Black an opening advantage. I'd prefer him starting out with 1.f3 and 2.Kf2, if we meet.

The 1.Nf3 (1...Nf6) and 2.Ng1 (2...Ng8) opening, followed by a "normal" continuation, is not new. It was used a decade or more ago, with the purpose of destroying opening trees, created by chess software programs, who reacted to these games, with the above starting opening moves, as if a virus had attacked them. Later they were deleted from the databases.

Just saw a nice, article from last April at http://www.thechessdrum.net/newsbriefs/2005/NB_TheMatrix.html! I recommend reading it. I just did. There is even a photo of GM Nakamura and 2.Qh5 and some interesting comments.

Good night to all and thanks for the posts.

Just saw a nice, article from last April at http://www.thechessdrum.net/newsbriefs/2005/NB_TheMatrix.html! I recommend reading it. I just did. There is even a photo of GM Nakamura and 2.Qh5 and some interesting comments.

Also interesting reading about the early development of the Queen is at http://chess.about.com/cs/reference/qt/bltip007.htm

Good night to all and thanks for the posts.

In a "normal" game, white is thought to have an advantage, because he's up a tempo.
Ex.1: 1.d4 d5 and white to move.

In the Nf1-g3 line, White's position does not "promise Black an opening advantage" ?!?!
Ex2: 1.Nf3 d5 2.Ng1 and black to move.

White has not played d4 AND black has the move. Surely this is at least as great an advantage as white enjoyed in the first example?

I think nakamuras rook "sacrifice" is beautiful and inspired. After that, white has no chance at all. Simply the best plan for the position. Very strong.

Good morning, morning.
Sorry to disappoint you, but, according to modern theory, White has no advantage at all, even being up a tempo. Proof of this theory exists in any good opening book available on the market.

White with a extra tempo should be sufficient for at least +/= although of course that is not irrefutable.

According to databases, White scores about 54%. This is a statistical advantage, which is worth about 40 ELO rating points.

My guess is when both sides are prepared, 1. e4 e4 2. Qh5 might score less than 54%, at least at higher level.

If you want a reason, you might consider that, in most openings, getting the queen out very fast is not seen very often in general... and the concrete variations after 2. Qh5 seems not to amount to much precisely because other pieces aren't developped... while Black has now an easier game at the center. If it wasn't for the stunner effect, I think this is no more than one of the numerous ways to throw away the 40 rating points advantage - not that it would matter much if you are playing someone with similar strength.

Peace...

White winning 54% of games is not the same as saying that white has a 54% statistical advantage based on opening play. Furthermore, the statistic is irrelevant when discussing one's chance in a particular game. There is no such thing as having a 54% chance of succeeding in a given game by playing white. The chance for success is 50% regardless of which color one plays. Most games between good players are not decided in the opening.

Hotep,

Maliq

"Volokitin played in the Greek Team Championship last year. He didn't leave a good impression there."

Being his teammate in that same championship, I should say that though his overall result wasn't fascinating, our team had no complaints about his performance and his team spirit. In the first 2 critical rounds, he scored 1.5/2.0, we assured ourselves of remaining in the First League and informed Volokitin that from then on, he could play as he wishes without considering the team result.

It is common, that a player playing for the first time for a team usually does not display his best performance.

It's about the way one sits, concentrates on the board, at his opponent, that impresses, for example, me, the most, in the beginning.

At a previous tournament, Volokitin made some fantastic result, I suppose I had expected a repetition.

It is good to have a team spirit, a rare quality in chess.

Multi-million game statistics, combined with Fritz & Friends chess engines, have created a false feeling of "chess mastery" among roughly 75% of the chess population, compared to the previous possibly 1%.

Some said computers will spell the "death" of chess and chess opening theory, in practice they are aiming at the "destruction" of human masters of chess instead, who are asked too often to "prove" themselves against machines.

IM.FM. whatever it is Jovan Petrovic i would just like to refute a couple of minor points made by you a)Morozevich has never played 1 e4 c5 2 Na3? that was Zviaginsev. I know Moro has a reputation for strange openings but this is not one of his.
b) I wonder why the ? annotation given that Zviaginsev scored 2.0/3 against Grischuk, Khalifman and Dreev.

IM.FST.J.P.: White is certainly thought to have an advantage. Why else might one ever consider play to win with white and draw with black? Superstition? Even if the answer is simply that white's position is easier to play and requires less precision at the beginning, though it is objectively not better, then I believe that constitutes an advantage!

Still, I believe it IS objectively better. Look at recent top-level results in the Ruy, for instance.

Maliq: "The chance for success is 50% regardless of which color one plays. Most games between good players are not decided in the opening."

That the chance for success is 50% in every game one plays is unlikely. There are an awful lot of parameters to consider. Still, assuming all else is equal, I still think white is clearly better. And as for the importance of the opening, didn't you notice anything about Kamsky's play this WAZ?

I hope there will be no argument that white enjoys a statistical advantage, but just in case, from chessgames.com, which seems to restrict itself to games between strong GMs:

White wins 134,973 games (36.73%)
Black wins 97,619 games (26.56%)
134,903 games are drawn (36.71%)

I am not claiming that these statistics are exactly correct, but it is obvious that white enjoys a significant plus even after presumably colour-blind selection criteria.

I'm not sure what happened to this thread, but I think alot of this garbage should have gone to the message boards.

Guys from Patra, common! Volokitin's performance in the Greek League was terrible although he was the most overpriced player of the tournament!

Peace...

Again, the chance of success in a particular game, based on color alone, is 50%. Other factors decide who wins or loses, but there is no inherent advantage to playing white aside from the option of choosing which opening setups to avoid. (For example, I can play 1. d4 and thereby never see a Ruy Lopez setup.) One cannot simply look at a success rate and determine that the difference is that one side begins with an advantage over the other; more often, games are decided in the middlegame or endgame, and when they are decided in the opening, it is because one side has gone terribly wrong at a point at which there were definitely better options available. Therefore, the relationship between color played and game result is spurious.

Hotep,

Maliq

You write very authoritatively Maliq, considering its a load of ... White isnt better than Black? Since when? Havent you heard of Black equalising? Ask any GM about it..

"Again, the chance of success in a particular game, based on color alone, is 50%. "

Obviously not, since, based on dozen of thousands of games, when White sits at the table to play, about 54% of the time, White wins.

"...but there is no inherent advantage to playing white aside from the option of choosing which opening setups to avoid."

This is actually an interesting theory, but in my small database of 1000 computers games, white scores exactly 54.4%. So it seems White, on average, avoids bad positions, rather than just setups he doesn't like.

"more often, games are decided in the middlegame or endgame,.. "

Yes and that's why White score is not 80%, 95% or 99.99%

Apparently Maliq's point has been missed. He is talking about a mathematical probability of either white or black winning... it is 50% (or 33% when considering a draw is probable). That's all. What this post is raising is another issue.

You can't merely look at game results without looking at many other parameters. A major factor may be the rating difference between players. BTW, chessgames.com does not limit its database to strong GMs, but take the example below:

White: player A (2700) Black: player B (2500)
White: player A (2500) Black: player B (2700)

Would white be better in both cases? Technically white's chances of winning is still 33% (win, draw, loss), but with the ratings there is another probability which changes with the rating difference.

How does one account for IM Manuel Leon-Hoyos (2428)of Mexico crushing GM Ivan Cheparinov (2625) of Bulgaria with the black pieces? Was Leon-Hoyos more comfortable because the tournament is in Mexico? All these factors are important, but are hard to measure.

Again... it is not clear that moving first is going to lead to a better result unless you take into account many other factors. Maybe the fact that when we have white, it makes us obligated to win. Who knows?

If it were as simple as the statistics given by "morning," then chess would not be a fair game (10% difference is substantial). We'd end up having to flip a coin for who moves first. Maybe that day is coming.

Peace...

d and zarghev, what you are doing is making a mistake in how to interpret the statistics that you are seeing. When one looks at a statistic that tells us that an outcome is expected 54.4% of the time, this statistic tells us NOTHING about the individual case before us. This is well-known. What it DOES tell us is that if 1000 games were played simultaneously, we could hypothetically expect 544 of these games to be decided in white's favor if all results are decisive, or that we can expect a combination of wins and draws to yield 544 points for white. Now, go ahead and try to predict the outcome of a particular game using this model, and you will immediately realize that it is quite impractical.

d, the idea of black equalizing simply is a statement that implies that the side to move cannot immediately create a favorable imbalance with this move. Because this is what is actually meant by equality, it is actually questionable whether white starts off with this ability to create favorable imbalance at all. It has generally been accepted a priori that white starts off with an advantage, and this conventional wisdom has been passed, virtually unchallenged, to current generations of chess players. The fact that GMs accept things as a given does not mean that they are so. Note how many pieces of conventional wisdom have been exposed by the power of raw calculation, and then do not imagine that current views must necessarily hold constant in the future. Your rude tone is both unnecessary and misguided, because you are not addressing a novice, and you would do well to discard it.

Hotep,

Maliq

Maliq,

Your outlook on the issue reminds me of the well-known joke:

"What are the odds of meeting an alien on the street?"
"Fifty percent, of course! Either you'll meet him or you won't!"

Peace...

I appreciate the humorous nature of your joke, Alex! Thanks for that! :-) Of course, the circumstances are different, because there is no definitive reason to imagine that one actually can meet an alien, both because we don't know with certainty that they exist and even their existence would not make it inherently likely that they could exist on earth and walk the streets. With regards to chess, we have a competition with a small finite number of outcomes, and each outcome is equally likely in a competition between competitors of more-or-less equal capabilities.

Hotep,

Maliq

Maliq I have to say that as I understand your point, and perhaps I don't, you are just wrong. It is best to look at it from expectation. Suppose you play white against someone with the exact same rating as you, your expected score from that game is roughly .54 (higher if you are exceptionally strong because this expectation must tend towards 50% as the players become weaker). In fact each outcome is not equally likely. To see this, suppose you are observing the battle of two nameless 2700 players. I bet on draw. If I can't do that, I'll gladly bet that white wins.

Note that if black is Anand and white is Shirov I'll take black. Color advantage is just one factor in the incredibly complex dynamic that is a chess game. You are right that strength and other individual factors like color preferences are more important than who has what color. But, it is impossible to deny that on average being white is advantage. This implies that if your job was betting on chess games and you knew nothing about the players, you would be a fool not to take the one who was white every single time.

"Apparently Maliq's point has been missed. He is talking about a mathematical probability of either white or black winning... it is 50% (or 33% when considering a draw is probable)."

According to this reasonning, the "mathematical" probability of getting a total of 2 by throwing two dices, is 1/11 (because the result is between 2 and 12, included). Needless to say, this is quite wrong and the mathematical probability is 1/36. Probabilities 101. What you defined has nothing to do at all with mathematics.
If you don't agree, I'm quite interested to met you urgently, to play all kind of dice games for money (or at least Poker, or Backgammon).

For chess, in fact, mathematics are totally unable to say what is the White winning probabilities in the game of chess, with imperfect (either humane or computerish) play.

However, what mathematics say, is that you count the number of times White won, you count the number of times Black won, you count the number of draws, on, say, one million of games, and then you can deduce some win/loss/draw expectancy with a good precision - if the million of games is not biased.

Then, you can say "if I take a random game, chosen in the same way as the sample, then White will get such and such probability to win/loss/draw".

If your sample is one million GM games, then it would apply to random GM games. If it is patzer games, or computer games, then it applies to them.

Now, the score expectancy of 54-55% is a fact.

Now, you can give theories and hypothesis to explain this fact:
- 1) White has an advantage - he has more chances to get a good position, with same level of play.
- 2) The games were not properly chosen statistically, for instance, they are all games where White is, in fact, Kasparov.
- 3) The games were not properly chosen, they are games taken from simultaneous exhibitions.
- 4) The strongest player takes White more often.
- 5) The Black players are almost always playing while drunk, while the White players do not.
- 6) The Black players feels overwhelmed, ... or White feel like they should win.
- 7) Players actually can't play opening well, they just play what other people play.
- ....

For now, a commonly accepted theory is that the reason for the score discrepancy is "1)", which is endorsed by most people, including, most of the people having some chess knowledge.

Now we can discuss calmly alternatives:
. "3)" for instance would be a very valid observation - because the player giving the simul is White more often than not. However it is not true that ~4% of the games in the databases are from simul.
. Likewise if you want to argue "4)" for instance you have the burden of the proof. I don't think it would be true in chess databases. On contrary, in tournaments, swiss or round-robin, the pairing is such that the stronger player has equal chances to play White or Black.
. And for "6)", it is wrong because chess programs score the same.
. and so ...

For now, I have yet to see any decent alternate explanation other than "1)". Feel free to bring facts to explain facts.

Peace...

Zarghev, we are not arguing with regard to what the statistics state. We are debating the interpretation of these statistics. The first thing that you must recognize is that any use of statistics must have a meaningful sample size in order to be relevant, and this is so because the resultant number only holds significance as an average, NOT as a predictor of an individual case. Given, for example, the statistic that the average black person has .919 years of education less than the average white person in a given sample, one should reasonably expect that a survey of all members of the population from which the sample is derived will yield a similar discrepancy: black individuals will, on average, have almost one year less of education than will white individuals. Now, if I was to take this statistic and apply it to a comparison between individuals, I would be hard-pressed to claim that Person A likely has one year less of education than Person B just based upon race. The reason is that those statistics are dependant upon holding all other influences constant, which cannot be done easily, if at all, in the natural world. Likewise, the statistic regarding the winning percentage for each color is only relevant if one can establish that this difference exists when holding all other things constant, which cannot be done. Owing to this, it is not sensible to bet that white will win a game 54% of the time simply because the player has the white pieces, because there are just too many other variables that figure into the equation for an individual case to be predicted so. The likelihood of a spurious relationship, or a correlation that appears to exist but is instead only the result of a relationship to some third-party variable, is far too great for one to use winning percentage as a viable tool in this argument.

Hotep,

Maliq

The question you seem to be forgetting is "what are we trying to prove?" A) That you can predict the result of a given game by using database statistics? Or B) that white has a significant advantage conferred by making the first move? It seems most people are saying B while Maliq is defending against A.

While I would still hold that White's advantage is significant enough to make a case for A, it isn't enough to compensate for other factors in a given game. But this is a strawman since I don't see anyone claiming that.

The question is whether or not, all other factors being neutral or, since that isn't possible, rendered irrelevant via a massive statistical sampling, having the first move is an advantage. A way to avoid having to analyze the particular circumstances around every single game (weather, hair color, other such apparently critical info) is to have a massive sample. With each additional game in the sample the relevance of other factors affecting the result drops slightly because we assume those affect both sides equally, on the balance. (Black will be rated higher the same number of games White will. White will be ill the same number of times. Black will blunder as often. Etc.)

That's what margin of error is for. With ten games it would be huge. With 100, medium-low. With 1,000 very low. That white scores 54.8% of points in a sample of 110,000 games between 2500+ rated players is quite relevant, unless you want to make the fantastic case that there exist extraneous factors that favor white in a way other than having the first move.

Why the first move is an advantage is an entirely separate issue. (Preparation, initiative, system accommating of style, etc.) It is one, clearly, demonstrably. Ask anyone, including yourselves, which color you would rather have in a money game. Again, not knowing your opponent. Of course case-by-case factors are more interesting and often more relevant, but having white is a big one in any given game. That it can be offset by other things doesn't mean it doesn't exist.

The problem is that people extrapolate this basic, and not particularly useful, fact into spurious statistical conclusions about things like specific opening lines. I.e. saying that 1.d4 is better than 1.e4 or something silly like that, or that X line of the Najdorf is better than Y line because from 200 games, one scored 58 percent and the other scored 48. One move could change the evaluation of a specific line entirely. And opening choices are about personal style and preparation, at least at the GM level.

This means that, despite massive statistics, it doesn't help to say that d4 (56%) is better than e4 (55%). (Better for whom is the logical rejoinder.) But it CAN be said that white is better than black, because style is not reflected. You can play passively or actively with either color, for example.

DP,

Are you saying white will win the majority of the games (if conditions being equal)? Given your stance, I don't understand how you state that you would bet that two nameless 2700s would fight to a draw. How does your theory hold?

In your second post, I'm not sure why you'd want to place a bet on white if you know nothing about the players. If your bet is that white would win, then you have only a 33.33% chance of being right. Bad bet. If black turns out to be the stronger player your probability is even lower.

All,

Would alternating which color moves first change the equation??

Looks like a good tournament with interesting games.

FYI - while Zvjagintsev originated 2. Na3!? (Although a gentleman named Ronny Bublitz played it in 1987, and an Italian gentleman named Luca Tirrito has played it a few times in serious events), people are referring to Svidler and Morozevich playing 2. Na3!? in the recent blitz event in Russia.

zarghev:

What educated person would say 1/11?? We've taken statistics and know that rolling a two on dice is 1/6 * 1/6 = 1/36 with the possibility of getting a "1" on each six-sided die. This has nothing to do with what I'm suggesting.

If you flip a evenly-weighted coin 10 times and get nine heads and one tail, what is the probability on the 11th flip to get a tail? Does the probability change from 1/2 just because you've gotten more heads?

I'm simply pointing out that there are three results in chess: win, draw, loss. Your mathematical chance of white winning a given chess game is simply 1/3... you can win with white, you can draw with white, you can lose with white. If you exclude draws, you can win with white, you can lose with white... 1/2. Probability does not change.

Now... you can look at 2 million chess results, but the probability above is still the same. However, if you are saying white is winning 54% of the time merely because they moved first, you will have a problem proving it. Chess is not that simple.

Probability is based on POSSIBLE outcomes, not EVENTUAL outcomes. That 54% figure is only good for saying how white HAS fared, not how white WILL fare in the next game unless you test for extraneous factors mentioned in previous posts. That's why it is difficult to bet on chess... too many other factors involved. Maybe we haven't caught up with theory for black yet... so many factors!

If you want to argue all the other salient factors such as strength, ELO difference, playing styles with either color, home-field advantage, level of trainers, openings, move order, data mining of past results between 2500+ GMs, then that's a topic for a Ph.D. dissertation.

Interested Maliq? (smile)

Peas

Maliq's stance may be arguing some obvious side-issue (that the specific players playing each side of a game has a much more dominant effect of determining who will win over whether a player starts with White or Black), but obfuscates the initial issue under discussion (whether White has an advantage overall, with no other data points to consider). Of course if Kasparov is playing Black, I don't expect to score 54% with white. Does Maliq's point really need to be made? The fact that he ties it into a discussion that is irrelevent with his point is why there is such contention.

He then shoots himself though with statements like "The chance for success is 50% regardless of which color one plays." "Again, the chance of success in a particular game, based on color alone, is 50%. Other factors decide who wins or loses, but there is no inherent advantage to playing white aside from the option of choosing which opening setups to avoid." which shows a lack of understanding of both statistics and his OWN point. Why does he think that winning with White or Black falls on exactly 50%, when there is a distinct difference between the two sides (White moves first)? That would be an extremely unlikely conclusion for any game where there is a vast complexity and a tactical imbalance. It's like saying that Tic-tac-toe must be a 50-50 game (when moving first can be proved to win/tie at worst with best strategy). This is not to say that in games such as these, moving first = automatic advantage. If the very large sampling determines that Black wins more often, that is an acceptable result as well, as is that the game in question IS truly even, regardless of who moves first. But the sample is large enough in chess to be relevent. White has come out on top. Accept it.

I think Mig is trying to be generous in explaining Maliq's position, but it is either super-obvious if Maliq is saying other factors are more relevant in determining who wins in chess, or plain wrong if he thinks that chess MUST be a 50-50 game at the outset. He also dismissed the predictive power of statistics and how it DOES apply to each and every game played that is included in the sample size (with great variance of course for any single game, which a statistical model accounts for) to a degree that suggests he doesn't understand the issue under discussion at all, or how to create and use mathematical models. Even with great variance for any single result, the reason why statistics are so accurate with a very large data set to draw from is because each single game points to the statistical truth overall, and contributes directly to it.

Motel,

Stern

Peace...

Stern, your arrogant reply serves both to provide a window into your character (antagonistic toward those who are not antagonistic toward you) and a failure to comprehend the statistical model that I was even discussing. My point with regard to statistics is that one cannot use database information to create a strong case for advantage of or disadvantage of starting the game with one color. I will be happy to give you a personal lesson on ordinary least-squares regression if you so desire, but please refrain from attempting to mock me and simultaneously disseminating misinformation.

Hotep,

Maliq

Daaim, why are you talking coin flips, or anything else that is demonstrably 50-50, when chess is demonstrably 55-45? Why conflate possible outcome (obviously three, win lose draw) with probable outcome as determined by statistics? Color in chess IS predictive, the data set is simply too huge to ignore or blame on other things. Take 30,000 games between players within a very tight rating range and it will come out that way.

Arguing that every single one of those games had a specific reason for its result - obvious fact - is exactly why we have statistics based on massive samples. They show trends and tendencies without having to analyze the specific games one at a time. In other words, what plausible explanation do you have for white scoring 55% from a sample of 110,000 games between players all rated 2500+? Coincidence isn't going to cut it with such a large sample.

Take your coin. If you flip it ten times and get 7-3, that's a coincidence, perhaps. It's too small a data set to be sure. If you flip it ten thousand times and get 7000-3000, that's a very unbalanced coin. If you get 5500-4500 that's a slightly unbalanced coin.

It's not entirely correct to say this is an apples and oranges argument. Having white is an advantage in a specific, given, game. It can be offset or increased, but it is a tangible factor.

If you flip a coin, there are three possibilities -- land on head, land on tail, or land on its edge. If some of the arguments above are true, then the probability of each of these outcomes is one-third.

We can agree that their are more salient factors that will have an impact on the probability of winning a given game (and Jeff Sonas has explored a number of them). ELO rating is one of the first.

However, that doesn't change the fact it is possible to make statistics on games in general, because, people take great lengths at giving White or Black side more or less at random - this is true for most tournament systems.

So let's again make the reasonning: let's take the results of some computer tournaments, where effectively an attempt is made to give White and Black at random (or better, alternatively), for instance, this:

http://www.husvankempen.de/nunn/cegtrating4040best.html

One sees (at the bottom):
"Games : 75359 (finished)
[...]
White Perf. : 54.5 %
Black Perf. : 45.5 %"

Hence the conclusion that White has an advantage for some reason. Now there might be a better explanation for the difference, but the burden of proof is on whoever claims there is another explanation.

The fact that White is chosen at random or alternated is important here ; as Mig pointed out, this cannot be applied to individual opening statistics, because a player does not choose his openings at random.

Lastly, yes, it is possible that the way to play and hence the statistics will change one day - indeed if ALL computers/humans were playing perfectly, the white expected score would be either 100%, 50% or 0% - not 54% [and more likely than not 50%]. For instance, maybe the Leko or Kramnik style is the right way, I don't know, one would have to look at performances.

"My point with regard to statistics is that one cannot use database information to create a strong case for advantage of or disadvantage of starting the game with one color."

Why not? Why is this not a good sample to draw from? Explain this, and maybe we can understand your point better.

"I will be happy to give you a personal lesson on ordinary least-squares regression if you so desire, but please refrain from attempting to mock me and simultaneously disseminating misinformation."

Knowing how to do the math and knowing what it means is a big difference. I will not lie - I do question your ability to draw conclusions from statistics because you are so dismissive of the large amount of data out there on chess. Even if you can do regression math. As to who is disseminating misinformation, that is still being determined.

Peace brother.

The outcome argument(draw,win,loss) is making the exact same flawed argument about the aliens(will or will not meet an alien) Now on the other hand it is not so easy from my perspective to formalize what is the correct approach to chess probabilities. The problem is that chess is not an inherently random thing(although when it is played by humans it is). Most likely the game has a correct outcome=draw(although win for white is also plausible). The 54% is the implicit expectation for white, when humans play chess. This is subject to change accross time of course and I guess this is what I understand Maliq to be saying. Imagine someone sitting down at the board has refuted the Najdorf or even figured out how to get a slight edge by force. I guarantee that the winning percentage for white will go up at least a little bit for a while. This argument has lead me to think about a separate point. Suppose that chess is a draw. Then we can expect that the more that the white winning percentage should go to 50% as people become stronger because there would be less "noise" in the system. But isn't it true that white scores better than 54%(I am not 100% sure of this fact please correct) at the very highest level (2700+). This suggests that 1) chess is not in the least played out and 2) that the game should eventually be shown to be a forced win for white. I think it would be quite interesting to study this a little bit, although probably rather mechanical for a thesis(although not necessarily).

Peace...

You are correct with regard to what I am saying, DP. All one can claim by assessing this 54% score for white is that white has scored this well in the games within the sample, and that one who analyzes the population from which this sample is taken can reasonably expect a similar distribution. This is not sufficient to claim that when two equally talented players sit down to play, one side has a 54% chance to win the game. If we were to do longitudinal analysis, we could derive a sample from any point in time and find a different percentage. As I mentioned in a previous post, there is both no basis for assuming that the mere fact that white moves first accounts for the 54% success rate (the influence of the white pieces may be more or less, but simple game results cannot tell us this), and no reason to imagine that this percentage would hold constant over time. If the percentage does not hold constant over time, then it is not an inherent advantage, only a notable trend that is time- and circumstance-specific.

Stern, you are quite correct that knowing how to do math and knowing how to interpret it are quite different. Being that I interpret statistics pretty much every day and work with others who have done so since before I was born, I will give an affirmative nod to your statement, but not for the reason that you might have imagined when you made your comment. At no point have I contended that the numbers quoted have no meaning; I have contended that they do not hold the meaning that is being applied to them.

Hotep,

Maliq

Easily the funniest thread ever on ChessNinja.

Still not answering the question about what the question is. The point, at least to me, isn't whether or not white is better in some hypothetical, infinite wonderland. That it has held constant to this point is a marvelous indicator that it will do so in the useful future (aka "time"). This isn't roulette. We are talking human results and predictive value in other human results, not eternity and perfect play.

We have anecdotal and common-sensical reason to believe the initiative of one move is worth something. This is also backed up by statistics to the tune of white scoring over 54% in games between evenly matched players.

So what are the other theories as to why white scores 54%, if it's not the first move? Why is that fact "mere" when there is nothing better readily available?

Mig,

What I am arguing is that white's advantage is not significant enough to attribute it to the first move. What do I attribute the color disparity to... lack of knowledge about chess. The advantage for white is only evidence of what we have found thus far from a small number of games played.

All,

I agree with zarghev last point. As we evolve in chess, will these numbers change? Yes! These statistics will certainly change because we have not explored enough possibilities in chess to say white is naturally better. If we say that two 2700s would most likely draw each other (with best play), then this is proof that white is not naturally better. Fortunately, chess has not been "solved" and we're still working on improvements.

Let's look at another intangible factor...

Thus far most chess literature has been designed to find ways for white to win, for white to attack, or for white to maintain initiative. You can even look in problem books. There are books where every single puzzle is "white to win."

There is something psychological about how we view the colors we play. We feel we SHOULD win with white. Is it because of the first move? No... it's merely because we have white. It's SUPPOSED to be an advantage. There is a psychological component that we do not realize. Some of this carries over in our attitudes over the board. Kramnik once said "a draw with Black is OK." Would Fischer say this? Of course not!

Have we ever heard that "with best play" chess games should end in a draw? Why do we say this? We are still learning what constitutes "best play."

The future...

Our knowledge is still limited in chess and like Morozevich said in a recent interview, no human really understands chess at this point. He was even so bold to say that "Kasparov doesn't understand anything in chess." Computer programs have shown us... for better and for worse.

Let me see if I have any understanding at all on this issue.

lets say I have lots of money and I have access to lots of chess games. and I am allowed to bet on them at no cost.

if the game comes up a draw I am returned my bet. if white or black wins the payoff is 1 for 1. I will always bet $1 on all my bets.

now I have access to all information about the game the set up the color the players ect. there are going to be 1,000,000 games and I am going to bet on all of them.

I am allowed to use any strategy I choose. I can look over a game and study it hard and make an educated guess on who to bet on. if Kasparov is black against a low rated player I can choose black if I want. I can develop any gambling scheme I want.

so now I am ready to bet and play this game. I have decided to always bet $1 on white in every game. I choose to not look at any other information.

I expect to get a score of 54% winning 46% losing. for a net 8% profit. on 1 million games I expect a net profit of $80,000.

now I have used only color to make my bets. nothing else.

Now Meliq are you saying that you expect me to make no net profit. that I will end up breaking even.

I believe that I will definitely make a profit on 1 million games betting on white in all games.

I dont need a theoritical foundation to make my bets. I am willing to go at it now without a theoritical foundation. I really dont care why any game wins or loses. or what different effects there are on a game. I am only interested in betting on white and winning money.

You can't mix phrases like "most likely draw" with the rest of that argument. Then you're back to probability and humans and useful statistics instead of the fascinatingly useless theories. We know an advantage is not the same thing as a decisive advantage.

As I said, WHY white is better is a separate argument. We have a vast wealth of experience that shows us a move has value. Suggesting that white wins because of psychology is odd when we find that computers have nearly identical the same results. And 100,000 is not a small number. It is huge. We know this because 1000 games, 10000 games, and 100000 games show the same results. It would be astonishing to see a five factor change after three factor accuracy.

As for why, as was said above, the burden of proof is on the person with a theory other than the first move. Otherwise we may as well say it's because the h1 square is white, or because having the king on your right is better feng shui.

Personally, I don't think the statistics of percentage of points will change significantly in human chess. Even with our much greater understanding today, things haven't changed since we had enough information to get a decent sample. (7000 games from 1910-20 show 55%.) Even if computers come close enough to solving the game, or reaching a near-perfect level of play, we KNOW how much the first move is worth in real chess today. And tomorrow. And the foreseeable future.

Back in the University (15 years ago) in the Abstract Algebra classes, we were given very nice example of a way to use the Fields theory in life: many of the board or number games, the games in which two players are operating sequentially, and there is no random or hiding factor involved (no cards, dice) were categorized into several groups. Games in which, in perfect play, the First To Move (FTM) will definitely win; games in which, in perfect play, the FTM will definitely lose; and a broad category of games in which if the two sides play perfect, the game will be drawn. And we proved on abstract level for many classes of games that they belong to one of these categories.

Somebody above said that with tic-tac-toe the "FTM always wins or ties, and the STM always ties or loses". Which was a wrong statement! At tic-tac-toe very quickly we learn that the game is always drawn with perfect play. Yes FTM has advantage and STM has to "react" at every move, but nevertheless the end result is a forced draw.

I believe there is still no existing mathematical model of chess that can categorize it into one of these groups. But it is not proven neither that chess is un-model-able. Simply we don't know enough yet. I believe that at some point there will be enough knowledge collected or a new algebra invented that will be powerful enough to model the game of chess into a more abstract category of games with proven outcome. And then we will learn that chess, if the two sides play perfectly, is most likely a forced draw (or a forced win for white, or a forced win for black). Of course this is sometime in the future.
And even if it gets proven that the outcome is determined, it is totally separate formidable task to learn "the perfect play".

My point at the end is that even though as of today white has 54% advantage statistically, over time this number will change towards 100%, 50%(most likely) or 0%.

There is no reason that should ever be true in play between humans, or between computers in the forseeable future. We know chess is theoretically limited. But it is vast enough that humans will never be able to remember the solution even if computer present a roadmap some 50 years down the line. That's why I don't think the stats in human play will change much. There will be a higher percentage of draws as technique continues to improve, but the draw percentage has more than doubled in top level play in the past hundred years without changing the percentage of points won by white.

This thread is yet another demonstration of the inability of folks to say: "I was wrong".

Mig asks: "So what are the other theories as to why white scores 54%, if it's not the first move? Why is that fact "mere" when there is nothing better readily available?"

Daim replies: "What I am arguing is that white's advantage is not significant enough to attribute it to the first move. What do I attribute the color disparity to... lack of knowledge about chess. The advantage for white is only evidence of what we have found thus far from a small number of games played."

Huh???
1. 5% is a significant advantage. In ELO terms, that's about 40 points.
2. Anyway, why should the advantage be "significant" to be attributed to the first move?
3. Why does "lack of Chess knowledge" skew the results in White's favour? Why not Black's? If the answer is, "White has the first move", I rest my case..
4. The sampling population used to derive 55% advantage is NOT small, by any statisical measure.

Mig,

We are speaking of computers as if they are perfect in their play. While they are near-perfect is certain aspects of play, they are also fed human data, so the results may be the same.

Svilen's comments are most incisive and as I've said before we do not know enough yet! It's simple. One hundred years (and 110,000 games) is not a lot of data unless we have exhausted the possibilities.

As we evolve, we will find that there will be points in history where black will experience a higher percentage of wins with more improvements. Over times they will even out... they have to. Otherwise, chess is a fixed game! Who wants to play a game where white has 10% higher probability to win?

Would it then be better to choose at random which color has the first move??

d,

Who worries about being wrong?

I just explained... most of the research we have found is skewed in finding advantages for white and not black... and we have seen the result of this in games played thus far. This has been the approach to the research. On the other hand, literature is written so that black "defends," "equalizes" or "fights back." We are looking at this wrong! Is black really try to equalize or win? If not trying to win, then what is black playing for? To draw?

We miss the psychological factors here and certainly one FEELS and approach chess as if white's first move is a deciding advantage to win the game. As I said, "white to win" books are prolific and even the boards are set up from a white vantage point in books and diagrams.

Heh. So you're calling this a chicken/egg situation? That overwhelming stats favoring white are due to a psychological perception that says white should be favored? There is no reason the move should be worthless and many reasons why it should have worth. That statistics back up this general impression and centuries of experimentation make the case overwhelming. You'll need some actual evidence to try and flip the entire thing around to "Barry Bonds has big biceps because he hits a lot of home-runs."

I never said computers are perfect. I said they don't have psychology. Play the same engine against itself with no opening book for a few thousand games and you'll get 55% for white. No human data, just the first move and a lot of boring games. Make a few uncontroversial opening moves and run the match from those positions.

Back when both sides played to win and draws made up less than 25% of results, it was still 55% for white. Bonus statistic: not counting draws, in decisive GM games of fewer than 26 moves, white scores 65% compared to 61% overall. In sharp contests (where you have more blunders and shorter decisive games) the move matters even more.

You've been playing a game where White scores 55% of the points since you first pushed a pawn, no big deal. The reason matches and tournament balance color distribution is because white has a significant advantage. That you say black will win more games in the year 3000 (or at some point before that) has little to do with this, especially since there is zero evidence to support this theory.

Look, don't respond so seriously to Maliq and Daaim, they just joking around .. I hope.

The point of the dice analogy is that it's something we know how to calculate. You can't say that because you know how to calculate the probability precisely for one case but not for another, then the logic is absurd for the first (the dice) but not the second (expected score for white). Just because we don't know how to calculate precisely the probabiliity of a whilte win, a draw, and a black win doesn't mean that the probabilities are all equal. That is absurd!

It is not inconsistent or unreasonable at all to say that, given a game between two players I know nothing about, I would be slightly more surprised by a 0-1 result than a 1-0 result. Not much, but a little. Of course, I wouldn't bet on a player 200 points lower rated to win very often. But if I know nothing else, I would expect white to win a little more than black to win.

"I just explained... most of the research we have found is skewed in finding advantages for white and not black... and we have seen the result of this in games played thus far. This has been the approach to the research."

I disagree - the kind of research Mig and others describe is color-blind, pun intended. Dip into any database and you see the same result: a 54%-55% expected outcome for white.

I also deal with statistics every day in my job, and I still don't understand your point. Unlike coin-flipping, probabilities sometimes have to be determined empirically (at least until Mig's 200-ply computer finds that guaranteed win or draw). And in those cases you do have to try to eliminate confounding variables(like mismatches in player strength or a bias in the sampling), but I don't see that as an issue here, since all databases seem to point the same way.

I suspect given what you said that you understand the difference between correlation and causation, but you don't always have to know the causation for the correlation to be meaningful.

PS Let's hear it for those Young Masters!

And I foolishly thought this thread was about the Cuernavaca tournament, sigh! Ended up being a heated discussion on the interpretation of chess statistics.

Regards,
Francisco

Count your blessings, Francisco. At least the off-topic comments aren't about Bush.

Certainly jonas...

Causation and correlation are issues that can be brought up here, but I didn't want to go that route.

I understand that many variables are hard to control for, but of course there are many factors that could cause white to have a statistical advantage... at this point in history. Is it the first move... it's hard to say. Most people on this board believe that is the most important factor. I still believe that psychological factors may play an additional role in white's perceived superiority.

Question jonas... if we randomized which color moved first, should the data show a 50-50 correlation? It should be statistically 50-50. I'm just curious because I do believe in the future that a variant will be created where black moves first. Also what do the color statistics show in Fischer-Random?

I'm trying to be open-minded about all this but I just don't get it. It is well-established that you can gain various types of advantage over the board, including a gain in material, a gain in time (development/tempos), and a gain in position. Obviously having White does not give you a gain in material (material starts out even) or a gain in position (positions start out identical) but it does give you a gain in time. You get to move first and force the issue a little bit. Various strong players have attempted to quantify the value of a tempo, in terms of some fraction of a pawn. But one tempo (or one half-tempo, if you prefer) is clearly worth more than zero pawns.

And if you have an advantage of one tempo, and material and position are equal, then overall you have an advantage on the board, roughly equivalent to some fraction of a pawn. And if this is all true, then you would expect to see White (starting with a slight advantage) averaging more than a 50% score, and indeed that is what we see, across the board, always and forever (so far).

Now it may be that the value of that tempo changes a little bit, depending on who you are. For whatever reasons, in the games of Alexander Morozevich and in the games of Judit Polgar, the advantage of the white pieces is not as large, whereas the opposite is true for Viswanathan Anand's games and for Peter Svidler's games. It could even be said that there is no advantage of the white pieces when Morozevich is playing, and probably some people will take that and run with it, saying that he is playing the chess of the future or something. More likely it's something very specific about his openings. But I think it would be pretty silly to say that there is no advantage to moving first.

By the way, here is a very enlightening article on the evaluation of material imbalances in chess, including (at the end) an estimate of the value of one tempo as well as estimates of the Elo value of a material advantage:
http://mywebpages.comcast.net/danheisman/Articles/evaluation_of_material_imbalance.htm

Thanks for the link, Mr. Sonas. Fascinating article.

A caveat -- the article occasionally presents statistical evidence to back up conventional wisdom (e.g. that Q+N is worth slightly more than Q+B). The problem is that stronger players often seek out the supposedly superior material imbalance, whenever they, e.g., convert an advantage into a win. Thus, a 2500 converting a plus over a 2300 (which would happen more often than a 2300 converting a plus over a 2500) would seek out Q+N because that's what he's been told. This can skew the results slightly.

IIRC, John Watson investigated this very same issues, and when he adjusted his findings for rating, he found no advantage of Q+N over Q+B.

And thanks also for weighing in on what has been a completely silly debate from the very start (even if the responses to Maliq & co. have been interesting).

Adorjan would likely have something to add to this. I don't mean to stir the pot or probably what I have to say is not so interesting anyway but here goes. Thinking about Jeff Sonas' post, we can all think of a situation where a tempo has value -infinity(or whatever)---zugswang. We also know that in certain positions the tempo is worthless and in certain it is decisive(try fooling around with the copy cat game first playing an e4 line and then a b3 line and you can see what I mean). Playing the devil's advocate, we can say that sure---white goes first, defining his plan and black is now in a perfect position to react. In otherwords, played perfectly,the starting position could be some form of a mini-zugzwang, where the tempo is virtually worthless. Jeff's argument in other words does not convince me--- I agree that the tempo somehow appropriately averaged over the space of positions probably has a positive theoretically quanitifiable value(this is not entirely obvious, but plausible) and that in certain positions could be worth some fraction of a pawn or whatever or even more. But the question is what is the value of a tempo in the initial position. What do you all think?

Practical experience of many generations of gambit players shows that in a typical opening setup three tempi are approximately equal to one pawn.

Of course this statement deserves many qualifiers - what kind of tempi? (NxQ is worth a lot more than Nc3) Is it a central or wing pawn? Just how "approximate" is this approximation? And what on earth is a "typical" opening setup?

Still, 1 pawn = 3 tempi is the sort of generic guideline that will, on the average, more or less prove itself, as long as you don't try to mindlessly apply it in each and every situation.

Daaim thinks this is about 'black' and 'white' ... rather than 'who moves first' ... funny guy.

Daaim says:
I'm just curious because I do believe in the future that a variant will be created where black moves first.

abc,

I'm sorry, but you don't know me and I'm not sure why you are saying that. You think all my comments about probability are about what color the pieces are. Paint the chess pieces green and yellow for all I care. Most chess pieces are black and white the question still remains. Try adding something to the debate instead of casting aspersion about things you know nothing about.

Daaim,

You and Maliq have been filling up a lengthy thread with the worst rubbish by far ever written on chess or statistics, in my personal opinion. I don't know you, but I suspect you were not joking in your nonsense. You don't know me, so you don' t know what I know nothing about.

Instead of at some point backing off gracefully from your nonsensical support for Maliq and your own weird derviations from that, you add wonders of insight like:

"Question jonas... if we randomized which color moved first, should the data show a 50-50 correlation? It should be statistically 50-50. I'm just curious because I do believe in the future that a variant will be created where black moves first. "

"As we evolve, we will find that there will be points in history where black will experience a higher percentage of wins with more improvements. Over times they will even out... they have to. Otherwise, chess is a fixed game! Who wants to play a game where white has 10% higher probability to win?"

"If we say that two 2700s would most likely draw each other (with best play), then this is proof that white is not naturally better. Fortunately, chess has not been "solved" and we're still working on improvements. "

**** STAR POSTING ****
"I'm simply pointing out that there are three results in chess: win, draw, loss. Your mathematical chance of white winning a given chess game is simply 1/3... you can win with white, you can draw with white, you can lose with white. If you exclude draws, you can win with white, you can lose with white... 1/2. Probability does not change."

If the majority here is establishing that white has a 10% better chance of winning merely because he/she moves first (tempo), are we then playing in a fair contest? In most sports and games, one side gets to move first, get the ball first, attempt to score points first, but the chances of winning are yet approximately equal.

Of course in chess the position builds up over time and the objective is not amass points, but to gain an advantage (either materially or positionally) which will hopefully lead to a win. I will not believe a game as sophisticated as chess can be boiled down to such a simplistic formula... you move first, you get a 10% better chance to win.

I understand what some databases are showing, but I don't believe we can attribute it to the first move only. IF (only if) the first move is worth that much, then what needs to be done to make the contest an even 50-50 in chess in every contest? That's the question. I agree with DP in that Jeff Sonas' analysis is not very convincing... especially when he uses words like "forever" and then says "so far."

Draw odds for black is interesting and some have suggested that a draw be a win for black because of the apparent disparity. I wouldn't go this far, but if what you all are saying is true (again... I disagree) then something certainly needs to be done. Any go, draughts and shogi players here? Do they have the same color disparity? Why would these board games be any different in theory? Now if who moves first is randomized, everything should be 50-50 in every mathematical sense and would not hurt chess one bit. Each player would a 50% chance of moving FIRST in each game. In other words there is no color advantage.

Lastly, no one has attempted to address the definite bias toward the first player... the player with the white pieces. Diagrams in books, chess literature, problem books are primarily from a dominant vantage point of white winning, gaining the initiative and/or advantage. Only in helpmates does black move first! This certainly makes a difference in our perspective on which player has the advantage given a particular color.

Go Hikaru!

abc,

You're taking snippets of my statements bringing them horriblly from their original context.

It's clear to me that Maliq and I were debating about point A (mathematical probability of winning a chess game) and others were debating point B (probability of winning based on databases). Mig said this in the beginning. We then started debating about point B.

You have added nothing. If you don't like the discussion... then leave. Why do you bother?

Daaim,

Actually you added nothing, except some vacuous pseudo intellectualism. Context or not cannot excuse this kind of thing:

"I'm simply pointing out that there are three results in chess: win, draw, loss. Your mathematical chance of white winning a given chess game is simply 1/3... you can win with white, you can draw with white, you can lose with white. If you exclude draws, you can win with white, you can lose with white... 1/2. Probability does not change."

All the evidence and logic is behind chess NOT being a fair game, in the sense of unbiased each game whether you are white or black.

To randomize who plays first each game does not help, as then by chance people will get an unbalanced number of games of either color a lot of the time over a series of games, such as a tournament.

As tournaments therefore endeavor to give everyone equal color balance, there is no big problem.

By the way I LIKE the discussion and I have no intention of leaving at your behest. Reading some intelligent posts and a few pieces of nonsense is entertaining for enough for me, speaking as one who nothing about so many things.

Well, the problem is, the game of chess has to start with _someone_ making the first move... and a game of tennis has to start with _someone_ making a serve... etc.

Which is why we alternate serves in tennis and alternate colors in chess. Yes, a single game of chess is not a completely fair contest, being somewhat skewed toward white, but a match or a tournament are, overall, fair... why is it so hard to understand?

I'm not sure why the thread has turned so ugly, but it's an interesting discussion otherwise.

"Causation and correlation are issues that can be brought up here, but I didn't want to go that route."

But you did...you've consistently brought up the uncertainty in causation (is it psychological? etc.) to question the mathematical correlation. My point was that you don't always have to understand causation to believe the correlation, though it would be nice.

"Question jonas... if we randomized which color moved first, should the data show a 50-50 correlation? It should be statistically 50-50."

For humans it would mean reversing the way they think about openings, which might lead to some problems for black moving first. But in principle it would probably be 50-50...I'm not sure that proves anything, since that would be true whether white was at 54% or 50% long-term.

"Also what do the color statistics show in Fischer-Random?"

That's a fun question. I wonder if there's a database with that information.

Peace...

Abc, it doesn't take much effort to see that what I have contended is not rubbish at all, but rather just the practical questioning of the application of statistics. It is quite reasonable to concede accuracy after rigorous test rather than to grant it initially without contest; I am not a huge fan of normative knowledge. As I have said before, the first move may yield an advantage of more than 54% to white, or less, but we cannot deduce simply from looking at results that white starts the game with a 54% advantage because of the great number of games that are decided in the middlegame or endgame rather than the opening. Those results get figured in, too, and while Mig has argued that number of blunders, etc. should even out over time so that the adjusted means for both colors should still reflect such a discrepancy when all other factors are figured in, this is far from an exact approach to this issue. (I could never get a paper published by using such reasoning, for example.)

I weighed in on this discussion not in an attempt to establish with certainty that white has an advantage or no advantage from the first move, but rather to point out that the mere use of database statistics is not sufficient to prove this. People are forever looking at statistics and running off with them without properly assessing what they suggest and what they cannot, and this is what I was speaking to. If the first move is an inherent advantage, then it is an advantage regardless of other factors, and then the other factors weigh in to either increase or reduce the advantage, which would give us our current 54%. What I pointed out was that it is an obvious mistake to look at the raw statistics and simply proclaim 'Oh, white has won 54% of the games, so therefore white starts the game with a 54% advantage."

In reference to the testing of this theory with computers, they may be the best recourse for such a test, provided that a couple of things are controlled for: a) Opening books should be turned off, because computers do not produce novelties, but rather follow known lines until they are out of their book; and b) The test should be conducted with the same program playing against itself, so that the only difference between the two is that Competitor 1 has white and Competitor 2 has black. In such a way, we rid ourselves of questions of fatigue, worry about paying the rent, fear, stereotype threat, and all of the psychological factors that Daaim refers to directly or indirectly which may or may not skew the results noticeably. We also avoid the problem of computers automatically playing into a line that may turn out to be unfavorable just because it was programmed into its opening book. I think that this would be a worthwhile approach. What do others say?

Hotep,

Maliq

What we know is that the one concrete factor we are sure all those 100,000 games have in common is that White moved first. Concrete factor, note. The move is a concrete thing in chess. That the player with white has an advantage because his king is on the right or because book diagrams are from white's perspective are not concrete.

When you have numbers so overwhelming it requires a refutation, not additional theories. If you want to introduce another factor you must prove it is concrete by controlling for the other factor/s (the first move, in this case). That you can concoct a few thousand alternative explanations, none of which benefit from any form of evidence, might be a fascinating exercise, but it stops there. If you can prove that players have trouble solving tactical problems from the black side, for example, you've done something.

Playing a program against itself without a book is usually incredibly repetitive. Entire games can be reproduced on occasion. There is no reason not to use multiple engines as long as all games are played with colors reversed. The Fritz interface can do this automatically.

For me, the best proof of the value of the first move is opening theory itself. A hundred years of intense analysis has revealed a white pull in line after line. And don't tell me guys like Kasparov and Fischer didn't spend as much, if not more, time on the black side. (Not referring to their results, which of course were fantastic even with black, but the work they did. Their improvements were universally available.)

I say I'm off for a walk around the block. Maybe I'll meet an alien. Or not.

Peace...

Mig, you are wrong. If you claim that the first move is the causal factor in the discrepancy, then it is not left to others to either accept this or establish that some other factor is to blame. You have to control for other factors in order to isolate the effect of what you consider to be the strongest causal factor. I deal with such questions everyday, so I know the approach that is required. Controlling for your own causal factor and then exploring all others just logically makes no sense.

Hotep,

Maliq

abc,

Let's be civil... no offense to you. I welcome a healthy debate and also welcome anything you have to say on the topic (whether I agree of disagree), but I don't appreciate someone merely stating that this comment is silly and that comment is silly. You sit back and enjoy the debate (which is fine), but if all you have to say is critical remarks that someone is silly, then it's better left to yourself.

Now...

You have used one quote of mine twice to try and trivialize all of my posts, but you didn't understand the context. There were two issues the floor: one of pure mathematical probability of a chess result (win, loss or draw) and pure mathematical probability of a chess result GIVEN previous games. These issues were being mixed.

One poster mentioned the probability of rolling dice and scoring a "two" at 1/36 probability. I then mentioned the probability of flipping a coin is constant even though you may have a majority of one result. My point was (which you didn't get) is just because white was won a majority of games in chess does not mean that the probability of a possible result changes. Two arguments are mixed here... that's where our troubles started.

On your comment... you made an interesting point which I considered. You have a tournament where who moves first is determined before each game. Obviously, two results are possible... white moves first, black moves first. However, the players still are paired with whatever color they are due, BUT who moves first is then determined by a randomizing system where each person gets an equal chance at that privilege. The idea is to take away any perceived advantages that either color may bring.

The one slight difference is that if black moves first the entire dynamics of the game changes since black's side is a mirror of white (king on left side).

I realize that it is a very, very radical change in the nomenclature of the game, but it is no more radical than Fischer Random... which I find very interesting.

Ok, I'm back from my walk around the block. Guess what - no aliens. However, I am not to be fooled by the empiric evidence about the absence of aliens. I refuse to mix the pure mathematical probability of meeting an alien on a walk with the pure mathematical probability of meeting an alien GIVEN previous walks. Let's not meet those issues.

Next time, I will walk around the block in an alternating clockwise/counterclockwise pattern. It's an interesting experiment, introducing various psychological factors, such as - looking for aliens around your LEFT shoulder can be significantly different from looking for them around your right shoulder. We'll see if I can find some correllation, not to mention causality.

"
"Any go, draughts and shogi players here?"

Yes, in go, the first player, Black, has an advantage of estimated to be around 6.5 to 7.5 points. This has changed over time, as better statistics on the game were made: http://senseis.xmp.net/?HistoryOfKomi - one or two centuries ago, Black didn't get compensation, and hence had a noticeable advantage.

"This certainly makes a difference in our perspective on which player has the advantage given a particular color."

Things like this make me wonder if you have ever played chess, or if you are above basic beginner level. The perspective that *I* have when playing chess, is I look on the board to see who is better, and only that.

Also you never answer to the refutation of your "psychological effect" ideas: the computers score 54-55% as well, and they are not subject to psychological effect.

zarghev...

"One or two centuries ago, Black didn't get compensation and hence had a notable advantage."

I don't understand that statement. How can black have an advantage without getting compensation?

Another lone statement out of context. Zarghev... you make ME wonder if you live in the same chess world. You mean to tell me that you absolutely have no psychological influences about chess and that you only look at the board and NO other factors? All I have to say to that is you're quite a special breed of chess player.

On the computer question... computers do not have psychology, but they are fed data by humans all of whom have inherent bias (good or bad) about the way chess should be played and what the first move is worth. Computers are not creating knowledge on their own... not yet.

No one is saying that psychology is the ONLY factor is determining the result of a chess game. However, it may be a factor in influencing a player's mental outlook in playing the white pieces. They may drive harder for the point because they EXPECT to win. How do you see people upset because they didn't win with white or didn't get an advantage with white? We hear it ALL THE TIME.

In go, in order to counteract Black's advantage of the first move a few points are deducted from Black's score and added to White's in determining the result at the end of the game.

Again chicken and egg? Placebo effect? They expect to win because they have the first move, which is a slight advantage. They don't win because they expect to win because they expect to win because they expect to win. I can't imagine we've all been the victim of a massive hallucination perpetuated by the power of positive thinking. (That we expect to gain an advantage with white because that's what we've always been told and not because of experience, centuries of it and our own.) Not when there is so much evidence for the first move being worth something. Say, 54.5%.

Sure many people play harder to win with white. That doesn't mean they score more points. Overpressing leads to many defeats. Again, computer play resolves this. Still 55%. Which leaves the only remaining redoubt that opening theory is heavily in favor of white. Why this would be true I can't imagine. There are just as many books on defenses. It's not as if opening analysis is one-sided.

"I don't understand that statement. How can black have an advantage without getting compensation?"

It should be: "White didn't get compensation".
So yes, in the game of go, it is well publicized that Black has an advantage. A player like Shusaku ,at the times where compensation didn't exist, was known to be virtually invincible with Black.

"You mean to tell me that you absolutely have no psychological influences about chess and that you only look at the board and NO other factors?"

Not really, but I look at relevant factors. When I have a good position, active pieces, good pawn structures, no weakness, equal material, I don't tell myself: "ha now I have a good position... but wait! no! the color of my pieces is Black! so I must force the draw! now!", or similar nonsense.

If you don't have an easier game with White, tell us what are your openings, this would be interesting.

"On the computer question... computers do not have psychology, but they are fed data by humans all of whom have inherent bias (good or bad) about the way chess should be played"

Some of the successful chess engines are done by people who have no so much chess knowledge - and *nowhere* in the code you see: "if (my_side == White) then position_value += 0.5". The opening books are automatically generated, and most of the time, either automatically or manually adapted depending on the results of the programs - hence most programs are avoiding closed positions - and also all positions where they are statistically losing too much. So why, at the end of the day, didn't they remove all the variations where Black scores 43.5% on average?

Maliq: "What do others say?"

I say it is unnecessary, as it ignores the overwhelming evidence already available. I say that opening theory is an honest best-effort to play optimally by BOTH sides, so it does not need to be thrown out.

I also say that believing in your complete self-assurance in your correctness just because you use statistics in your everyday job and your using it as part of your argument is like watching a bad comedian who thinks he's really funny perform. It just makes me cringe. You don't work for the Accounting Department at Enron, do you?

As for Maliq's and Daaim's alternative explainations of brainwashing by books to favor White, opening theory developed to specifically help White, humans not being evolved enough yet to play Black correctly, "probabilities do not change" arguments, and 500,000 game databases being poor sources of data, etc., ...

VS

Having the first move in chess offers a small but tangible benefit...

I turn to:

Occam's Razor:

"Given two equally predictive theories, choose the simpler one - The simplest answer is usually the correct answer."

One should not increase, beyond what is necessary, the number of entities required to explain anything.

I'm pretty sure we are beyond the point of convincing anyone with an opposing viewpoint of anything different than what he chooses to believe. I take comfort in the fact that there is a reproducible, statistical pathway for those now and in the future to determine for themselves if White, with first move, has a tangible advantage in Chess.

OK... let's try this again.

zarghev,

Thanks for the "go" info. It's always been a fascinating game to me, but I've never learned to play.

No one suggests that you are driven by purely emotional decisions when looking at positions for black. However, it is a common thought that one feels more confident in playing the white pieces because they learn (through various ways) that it is supposed to be advantageous.

In terms of computers... certainly no one writes code to say "if color == white, then evaluation = +0.5". Computers don't work in that way and do not work without human intervention and input. Of course, data (from the 54.5% win percentage) are fed into computers and a host of other variables. Certain database tables, algorithms and search methods are used to arrive at a final eval. You know this. The point is that computers do not figure out anything but what is already an outgrowth of human data that it is fed. (Note: I would be interested in Mig's experiment... taking the computer databases offline.)

I'm confused... you asked why don't they discard the 43.5% games that Black won? Why would they do that? You meant discard the 56.5% they lost, yes? They could do that. They could do many things; they could improve hardware/software; they could run more tests; test could eliminate redundant searches. Why haven't they made these adjustments? I don't know. However, that doesn't mean that black is permanently relegated to 43.5% kill rate (or whatever the figure is).

Mig,

What I'm saying is that one's mental outlook can be affected by whether they perceive an advantage or disadvantage with a certain color. Most players, for whatever reasons, feel an advantage with white. Maybe it is the "white to play and win" books. Maybe it is the diagrams in 100% of the books that are viewed from white's vantage point. Maybe it is the 54.5% kill rate. Who knows?

You say (I'm paraphrasing), people may play harder, but that doesn't mean they score more points. Do you believe that a positive mental outlook has nothing to do with results? I'm not sure of your sports background, but I played organized sports (at a decent level) and still do martial arts. What are stating goes against everything I was taught and learned as a competitor.

I believe that players feel more confident with the white pieces because they understand (as you and others have conjectured) that it is an advantage. That mere motivation is worth something during play. However, statistically speaking, it would be a hard variable to measure because there are other factors at work. One could ask a direct question whether players feel more confident with the white pieces (or black) and then record their games result and do a correlation, but of course this doesn't prove causation because there are so many other factors involved during a chess game. (jonas... I went there).

As I said, psychology is NOT the only factor and I never suggested it was. I am merely bringing it up as a variable in explaining the percentage results that you all have given. Again... I don't believe chess is a simple as boiling the advantage merely to the first move... at least not to the clip of 10%.

stern,

I do appreciate and agree wholeheartedly with your conclusion.

Peace...

As I said before, Stern, overwhelming evidence that white has done 10% better is available; overwhelming evidence that white begins the game with a 10% advantage is not. All of this has only been to say that it is quite erroneous to simply look at winning percentage and conclude that this statistic can stand as a proxy for a measure of first-move advantage. The problems with such an amateurish conclusion are obvious. Furthermore, at which point did I ever talk of brainwashing, etc? At least know my argument if you are going to contest it. I have stayed quite close to arguments of the interpretations of statistics and have not strayed at all to matters of the metaphysical or anything similar. You apparently enjoy boxing with shadows.

Hotep,

Maliq

Talk of an objective "10% advantage" is silly anyway. The game of chess has *one* objective result, and it's either 1-0, 0-1, or 1/2 (assuming perfect play for both sides). We do not, however, know what that result is, since chess has not been solved yet.

Therefore, all we can talk about is a 10% *practical* advantage. And based on results so far, in games played both by masters and by computers, white has had a 10% practical advantage.

Therefore, Maliq is chasing a ridiculous chimera when he says that "evidence that white begins the game with a 10% advantage [is unavailable]." Of course it's unavailable, since white does not begin the game with any percentile advantage, except 100% (certain win), -100% (certain loss), or 0% (certain draw).

All we can do is quantify a practical advantage, which is inherently based on the skill levels of competitors and the state of chess knowledge at this time.

“As I said before, Stern, overwhelming evidence that white has done 10% better is available; overwhelming evidence that white begins the game with a 10% advantage is not. All of this has only been to say that it is quite erroneous to simply look at winning percentage and conclude that this statistic can stand as a proxy for a measure of first-move advantage. The problems with such an amateurish conclusion are obvious.”

I agree with this to some degree. I too have abstained from assigning a specific percentage number to White’s advantage. I think first-move advantage can be more pronounced or becomes more negligible in different cases (samples), particularly if there is a very large difference in elo rating or if both players are very low rated (say, elo 800) like a game between two 6-year-olds who just know how the pieces move but their tendency to leave pieces ‘en prise’ is enough to minimize any first-move advantage. But I will still maintain that there is enough (overwhelming) data to conclude that White has *some* tangible advantage due to the fact that he has first move, in all cases. Even for the case where Black has a much higher elo, the White advantage might manifest itself in White winning an extra game in 500 or 1000 over what he would do if colors alternated between games. It’s still an **advantage**.

And I think the statistics for games where players are 2500+ elo producing a ~54.5% expectation for White is accurate and applicable to predicting future results in this category due to the extremely large sample size.

My main problem with your posting Maliq was these three statements from you:

"The chance for success is 50% regardless of which color one plays."

"Again, the chance of success in a particular game, based on color alone, is 50%. Other factors decide who wins or loses, but there is no inherent advantage to playing white aside from the option of choosing which opening setups to avoid."

“"My point with regard to statistics is that one cannot use database information to create a strong case for advantage of or disadvantage of starting the game with one color."”

These three statements ignores all the statistical data on chess, and still makes me doubt your ability to use and interpret statistics. You think I’m wrong to contest this. I do not. Fine, we agree to disagree because I’m convinced neither of us is going to change our minds.

“Furthermore, at which point did I ever talk of brainwashing, etc? At least know my argument if you are going to contest it. I have stayed quite close to arguments of the interpretations of statistics and have not strayed at all to matters of the metaphysical or anything similar. You apparently enjoy boxing with shadows.”

I said “Maliq's and **Daaim's** alternative explanations.” To be fair to you, since some on the alternative explanations are truly out there, I probably should have separated the two of you, but who said what, and which on the list applies to which person is already clearly documented on this thread.

I’m truly done with this. Peas brother.

And if it wasn't crystal clear, the reason I think that the statistics show an advantage for White is due solely to White having the first move. No other explaination for the results we see over so many trails is logical, no other reason is needed as it is the simplest explaination and also the most obvious imbalance between the White side and the Black side in chess (so we still disagree on this point fully Maliq). Occam's razor.

Yeah, yeah, I know - "I thought you were truly done with this". I am now.

Maliq I dont whether you're a scientist, but in my opinion you show a fundamental lack of understanding about a theory and proof. You write:
"You have to control for other factors in order to isolate the effect of what you consider to be the strongest causal factor. I deal with such questions everyday, so I know the approach that is required. Controlling for your own causal factor and then exploring all others just logically makes no sense."

This is like the debating style of Mr. O Reilly.. I know I'm right, you're wrong, therefore I'm right. What other causal factors are you talking about?
1. The psychological advantage that White has because of the way opening books are authored?
2. The affinity White pieces have with white light?
3. Black gets to choose which side the clock is on, but since most players are right handed and choose the right side, this in reality favours White because then he can simultaneously record the move with his right hand and press the clock with his left hand?

Every single one of the potential reasons you and Daim have come up with have some trivial reason associated with them that precludes their being a candidate for the 10% advantage enjoyed by White.

Remember, a theory is just a model, to be proven or disproven by observations. In this case the model is, the first move confers an advantage. The proof is the 54% advantage shown by statistics. Does this mean it is impossible there is a reason other than the first move? No, but you have to come up with some alternate model. This is where the philosophical reasoning tool known as Occam's razor comes in. The duck test version goes like this. If you see something that looks like a duck, quacks like a duck, and even tastes like duck when you eat it, then in all likelihood it is a duck. It could also be a alien dressed up very cleverly to look, sound and taste like a duck, but we have no way of knowing this, and therefore disregard it as a viable model.

So if you come up with a model that for example predicts an advantage for White based on the three reasons above, then we'll certainly consider it. Otherwise its a waste of time.

I suspect that everybody other than Daim and Maliq will run out of patience with this thread and stop posting. I know I have..

d,

You're right. It's tiring... especially when you, stern and others mix up who said what and misconstrue the issues. The practice of taking one statement here and there and critiquing it is common in this medium, but is often abused as it has been here... taking statements out of context and then debating the distortions has been a common practice in this debate.

There has been so much twisting and adding to what I've said in my posts that it becomes fruitless to continue repeat the points over and over. When one starts adding jokes, hurling personal attacks and posting silly comments in a debate, there is no hope for a serious exchange. It shows that maybe we should have the "set kib" option for blogs as they do in the ICC.

I cannot speak for Maliq, but everyone is focused on the 54%. That is not the issue I was arguing... if the stats show that white scores 54% so be it. The argument is... what does the 54% represent? Is it a constant? Can it be improved for white? Can black improve? Does the first move result in the 54% kill rate for white or are their other factors at play? With a mere 100,000 games out of millions of "reasonable" possibilities for a chess game, we have yet to understand what this figure represents. Maybe you are set on that figure and that's fine.

Go back and read jonas' post, svilen's post and macuga's last post. These posts provide variety and balance in a debate whereas a few people have decided to hurl childish attacks against both me and Maliq for raising questions of inquiry. None of the personal attacks came up with any concrete analysis... only the repetitive "54% proves it." Stern just raised points that have already been debated about 20 posts ago (e.g., the rating element).

What if the bird you THINK is a duck is actually a goose?

ah daaim, you poor thing.. Because its not what I THINK old son. As for childish, do you have the absolute authority to decide what constitutes childish and what doesnt? And what constitutes variety and what doesnt? To me, your points are not only childish, they're asinine, ably demonstrated by your last line. Your stock answer to any point you cannot counter is to attack the tenor of the post...

d,

To my last comment... I'm just posing a question to your duck analogy. In other words, what if what we see something, but it is really a misinterpretation?

No offense was intended.

Hate to rehash this old argument but I think that the point of why white has an advantage has been missed. It has nothing to do with statistics, it has to do with basic understanding of the game of chess. One type of advantage in chess is time, seen in the form of development. With the first move I have the advantage of time. If I play 1.Nf3 there is no denying that I have a lead in development, which in itself is an advantage, period!!!

Wayne